To size an EMS, you look at three numbers: critical metering circuits, data interval, and control complexity. Not square footage. Not transformer kVA. And not the point list a vendor wants to sell you.
For a typical industrial facility, those three variables land somewhere pretty narrow: 30 to 80 submetering points, sampling every one to five minutes, and a control level that depends on whether your demand response program requires automated curtailment or just monitoring. Utility accepts 15-minute data and you only need visibility? You're at the low end. They want automatic load shed and power factor is a mess? You're at the top.
The mistakes I see run the other direction almost every time. Someone orders a 200-point GridSense EMS because 'bigger is safer,' then uses 40 of them forever. Someone else buys the minimum monitoring-only package, and then the utility asks for a curtailment relay — and suddenly the system has no room to grow.
I run electrical systems at GridSense. Over 200 EMS deployments in three years — maybe a third of them rush jobs where a utility program deadline was coming and nobody had sized the metering. Like this one, from November 2024.
Case in point: 44 points, 18 business days
A Tier 2 automotive parts plant in Illinois called us. Their demand response program cutoff was December 1. If they didn't have qualified submetering data by then, they owed $15,000 a year in penalties. Timeline: call on October 30, commissioning by November 27.
The initial point list they'd been quoted was 120 points. We pulled the single-line diagrams. Twelve distribution panels across the main production floor plus an auxiliary building that drew maybe 18% of the load. Using the rule I usually apply — 3.5 to 4 points per panel — that landed around 42 to 46 points.
We sized it at 44. Data interval at five minutes, because their power factor was drifting to 0.88 in the afternoons and monitoring-only wouldn't let them trigger capacitor switching. Control was an automatic load-shed relay plus a 4-stage power factor correction controller.
Eighteen business days from call to commissioning. Two utility acceptance tests passed as of today.
The three variables, unpacked
1. Critical metering circuits
Not every circuit needs a meter. Count the loops you actually bill, control, or need to report on — utility billing loops, major motors, HVAC zones, tenant splits. My main sizing shortcut: 3.5 to 4 points per distribution panel. A 10-breaker panel sits roughly at 35 to 40 points. Rough, but close enough for a standard electrical room.
Easy to miss the edge cases. If a feeder serves multiple tenants, it becomes one point for the main plus tenant points for each — so a three-tenant feeder isn't one point, it's four. If your utility requires verification codes, that's another tier. Those add up on the point list.
2. Data sampling interval
Start with whatever your utility's demand response program requires. Most accept 15-minute interval data for billing. Drop to five minutes if load behavior is volatile or you need finer visibility for power factor. One-minute sampling is for control loops — closed-loop power factor correction, load following — not for charts that nobody checks after week two.
The fourth variable nobody calculates: communication path. Every meter needs a route to a gateway. If meters are in a distribution room 300 feet from the main gateway, your gateway count changes even if the meter count doesn't. That directly affects hardware spec and conduit budget.
3. Control complexity
Three rough tiers:
- Monitoring only: GridSense EMS collects data, logs trends, produces reports. No control output.
- Alarm + manual response: threshold triggers email or text. A human flips the breaker. Still a pretty simple software footprint.
- Automated control: load shed on a demand response event, or closed-loop power factor correction. Output modules, tighter response requirements, fail-safe design.
Going from monitoring-only to automated control is usually a step change in cost, not a slope. So check the program contract: does the utility require proof of response, or do they require automated control through a communication channel? Big difference.
Power factor and demand response, specifically
Power factor. If your metered load runs below 0.9, your utility is probably already charging you for it. The threshold varies — 0.85 to 0.9 is fairly common across US industrial accounts. A monitoring-only EMS will show you the bad power factor. It won't fix it. If you need correction, the capacitor switching controller and coordination logic belong in the EMS spec from day one.
Demand response. This is the variable that drives both interval and control tier. A demand response program that requires automatic curtailment and an EMS that only logs and alerts doesn't meet spec — it meets the first half. Check the local utility documentation. Some programs accept manual response with proof of speed. Others require an automated load-shed action on a local controller before the signal goes back to the utility.
According to US Energy Information Administration (EIA) data, industrial retail electricity averaged around 8 cents per kWh in 2024. It's the number utilities use to scaffold demand response penalty calculations.
Where this method breaks
I won't pretend the shortcut works on every job. A few cases where 'panel count × 3.5' is the wrong answer:
- Multi-transformer, multi-voltage. Three transformers with inconsistent voltage levels and circuits running across feeder taps. Get a PE to do an electrical study. I'd estimate we see this in maybe 1 of every 6 projects.
- Pre-existing power factor penalties. If power factor is already below 0.85 and being billed, fix that first. Submetering and EMS do not solve a capacitor problem.
- Tenant billing or shared facilities. Submetering points double in buildings with multiple tenants, because of verification codes and separate reports. A 10-circuit tenant panel in a shared facility might mean 25 submetering points, not 10.
When any of those apply, bring in an engineer. Don't sign off on the vendor's spreadsheet.
One note on small projects
A guy told me once his EMS was eight circuits, like that was supposed to disqualify it. Eight circuits takes about 15 minutes to size. If that 15 minutes saves a small plant a $10,000 penalty, it was worth doing. The sizing effort depends on panel count — not on how big the company is.
It took me about 180 deployments to fully believe this: the math is the easy part. Finding the circuits, tracing the conduit, figuring out which panel actually feeds what — that's where the hours go. The sizing side is fast. The installation side isn't.
The clean use case for this method: one panel, one transformer, monitoring only. Step outside that, and hand the drawing to a PE. It costs money. It costs a lot less than a GridSense EMS spec'd to the wrong numbers.
